Air France 447 Black Boxes Readable

An anonymous reader writes "It's not a lengthy press release, but it's good news: the memory cards for the flight data and cockpit voice recorders from the Air France 447 crash in 2009, recently recovered from the sea floor almost two years later, are readable. The data was recovered over the weekend and includes the full two hours of cockpit recording. Apparently it will take weeks for analysis of the data, but it looks like the challenging recovery effort is paying off in a big way. Hopefully detailed answers about the cause of the crash will follow."

Comments

Yey for solid-state memory!

[KreAture]

Tragic story so far, but atleast it shows the viability of solid-state memory. On a sidenote: If there is only 2 hours of voice recording, why will it take weeks to listen to it?

Re:Yey for solid-state memory!

[Anonymous Coward]

It's probably in French.

Re:Yey for solid-state memory!

[JustinRLynn]

They're not just listening to it. There are a lot of recovery techniques that must be performed very carefully and with strict controls to recover every last bit of intelligible information from those media.

Re:Yey for solid-state memory!

[sycodon]

Probably an endless stream of cussing towards Airbus and "that damned computer".

The only wires planes should fly by should be steel wires.

Re:Yey for solid-state memory!

[Wyatt Earp]

Two hours voice recording, data logs from a few hundred sensors, it all has to be done meticulously and perfectly in accordance with aviation authorities from Brazil, the UK, France, Germany and the United States. And lawyers, alot of lawyers.

"The download was completed in the presence of two Brazilian investigators of the Aeronautical Accidents Investigation and Prevention Center (CENIPA), two British investigators of the Air Accidents Investigation Branch (AAIB), two German investigators of the German Federal Bureau of Aircraft Accidents Investigation (BFU), one American investigator of the National Transportation Safety Board (NTSB), an officer of the French judicial police, and a court expert. The entire download was filmed and recorded."

http://www.bea.aero/en/enquetes/flight.af.447/info16may2011.en.php

Re:Yey for solid-state memory!

[Luckyo]

Tell that to B-2 pilots.

Actually scratch that. Tell that to ANY modern pilot, be he military or civilian. For added bonus, tell that to greens all over the world and be lynched on the spot, as unstable aircraft are significantly more fuel efficient and can only be flown with fly-by-wire. Trying to fly it manually will result in very spectacular and fiery ending.

Re:Yey for solid-state memory!

[theshowmecanuck]

This is a very important point. The voice recordings will not make sense or give an accurate picture of what happened without being put in the proper context. In this case, the context of what was happening in the flight and what the aircraft told them about it in synchronization with the voice recordings. The voice recording on its own *could* give a completely misleading view of what happened otherwise.

I'm curious of whether this will help convict or exonerate Air Bus on their manslaughter charges.

Re:Yey for solid-state memory!

[UnknowingFool]

Most likely the investigators will not be listening to it like most people listen to a news broadcast. They will repeatedly be going over every second of the voice recording analyzing every pop, hiss, bang, etc. They will be analyzing what the pilots are saying and how they are saying it. (Are they stressed? Are they relaxed?). They will also been matching the audio track with the data track in terms of timeline and looking for any clues that the verbal or nonverbal sounds can provide. i.e. At 1:34:42, the data records that the pilots tried to pull the plane up while the voice records what may be considered grunting. Are the pilots exerting themselves in trying to control the plane? If so, what would cause the controls to fail like that? During the entire time, the investigators have to document everything including a transcript of events.

Every little clue may help in determining the final cause. In Continental Flight 3407, investigators thought they heard yawning from both pilots during the flight. The co-pilot had commuted from Seattle to Newark the night before possibly getting as little as 4 hours of sleep and appeared to be under the weather. In the end, the crash was attributed to pilot error brought on by a number of factors including poor pilot training, fatigue, etc. It also brought the issue of the long distances that some pilots have to commute in order to do their jobs.

Re:Yey for solid-state memory!

[JWSmythe]

    Trust me, it's not modern airline pilots who are screaming that.

    Not to make light with a car analogy, but.... :)

  I was driving out to watch the shuttle launch this weekend (yes, it was cool, even though there were low clouds). About 45 minutes into the trip, my truck stalled. It took a few tries, and it started back up. I happened to have my OBD-II reader with me, so I stopped at the next exit and checked for codes. No current codes, but a pending code for an intermittent fault with the crank position sensor. Hrm. It hadn't done that before. I stopped by a local parts store (found via GPS, not by knowing the shitty area), grabbed one, and continued on the trip. The car was acting fine for over an hour, and suddenly nothing.

    So just over an hour, and just about 100 miles from home, the engine cut out, and wouldn't restart. 5 minutes of finding the right tool in the back (I travel prepared), and 2 minutes changing the parts, and I was on my way.

    Cruising along on the Interstate, everything was fine. When we got close to Titusville, it started shifting funny, and the speedometer became erratic at anything below highway speeds. No codes, no pending codes, nothing. The common solution, the vehicle speed sensor (VSS). There happen to be two VSS sensors (transmission and transfer case). Two VSS sensors later, and all is fine again.

    The computer uses the VSS, TPS, and a few other things to determine what gear the transmission should be in. Cruising along at 45mph it'll up and down shift as it sees fit, which is much less than ideal.

    So manually shifting the automatic to 3rd gear to drive at 45mph or below, and shifting to 4/OD to drive above 55 was much less than ideal.

    There's nothing like fucked up sensor to ruin your day. If I hadn't been lucky enough to catch the pending code and pick up the crank position sensor, I would have been towed to a shop would would have been very happy to overcharge me for changing it.

    The engine is based on one with a great lineage, and the older pre-computerized parts are interchangeable, and the swap would be about $300 in parts. That would take care of any engine stopping problems. I haven't found a compatible transmission option that fit with the transfer case and drive shaft lengths. {sigh}.

    In my case, a faulty sensor or three just made the trip difficult. In the case of an airliner, ... well ... non-running engines, or inability to receive accurate airspeed readings are rather catastrophic. In the case of computer *ASSISTED* systems, the pilot can fly by the seat of his pants. Great, so the IAS is 0 knots. Big deal. We see clouds going by, the stick isn't shaking, and the physical stall horns aren't sounding. IAS of 999 knots means throttle back, and pull up to bleed some speed off. If that IAS is wrong, you're going to discover what it was like to be the apple falling at Newton's head.

    The car with failed sensors like this doesn't make the news. Well, not any more than me bitching about it. In a commercial aircraft, it'll make the news for years, a fortune being paid out to the victim's families.

    I'm not trying to argue that my car is more important than an airliner, but we see improperly implemented technology does cause lots of problems, which cannot be resolved by any degree of operator skill.

Re:Yey for solid-state memory!

[KreAture]

The sensors are actually very good.
The connectors used to interface with the sensors are however the cheapest you can get that has a vibration-rating.
It's sad, but there are connectors that will last the life of a car, but they will most likely never be found in one due to cost. Most car-connectors aren't even properly dust and mosture-proofed.

Re:Yey for solid-state memory!

[JWSmythe]

    Well, most automotive sensors I've seen for essential purposes (like engine and drivetrain monitoring and controls), are "Weatherpack" connectors, which do an excellent job. I don't know how the sensors I replaced were damaged, but it was physically obvious on the interior portion. I haven't seen a failed weatherpack connector yet. I have seem some that weren't attached correctly (stuffed in "good enough", but not electrically sound).

    Most aviation connections that I've seen are screw on connecters. That is, they push into place, and then the ring is tightened to secure it in place. I don't do much with aviation, but I've observed these used quite frequently.

    Neither takes into account stupidity of people who think they know what they're doing. I've seen duct-tape wrapped wires (hint: conductive), wires, and wires just twisted together and jammed under the carpet. Those always make for fun diagnosing problems. For some reason, I swear every friend of mine that has bought a used car, gets one that someone had put an amp and sub in the rear, and never thought out the design. I was helping a friend with his SUV, and found a huge (2 AWG, I believe) wire screwed to the positive battery terminal with a sheet metal screw, passed through a hole with no grommet (but yes, sharp metal), snaked through the cabin, finally ending under the carpet by the rear doors. The last 4" were stripped (why? I don't know), and it was precariously close to bare metal. Well, precariously as in there was evidence of arcing. There was no fuse anywhere from the batter to the end of the wire.

    Sometimes I wish they didn't sell all the parts for any idiot to play with their wiring. Then again, it would make it more difficult for me to get parts and set things up correctly. :)

Re:Yey for solid-state memory!

[Cochonou]

Actually, there are already reports saying that according to "leaked" information from the investigators, after preliminary analysis of the FDR contents, the likely cause of the crash is not an aircraft malfunction. See this very rough google translation. Of course, such early reports should be taken with a lot of caution, but they come from a rather respectable source.

Re:Yey for solid-state memory!

[adolf]

I had a Weatherpak connector (not quoted, because I believe it actually was a Weatherpak(tm) and not just something similar) fail: The connector on the coolant temperature sensor on my work truck broke into little pieces when I was changing the spark plug wires, just a few weeks ago.

Of course, changing plug wires is one of the lowest-impact maintenance events that ever happens on a vehicle. If I'd broken it doing something else (like wrenching out a stuck spark plug with a breaker bar, two u-joints, various extensions, and grunt, and a blackened thumbnail), that would seem reasonable, but this wasn't one of those situations.

Further investigation revealed that that the sensor-side of the connector was very brittle, and bits would break off easily in my fingers: One might say it had intrinsically failed all on its own long before I nudged it into disintegration: If it hadn't been nudged by me, it would've nudged itself eventually. Fortunately, water seemed to have mostly stayed out and the cable-end of the connector seemed OK, but if I hadn't nudged it into complete failure at that time it might have festered into a more difficult repair.

It was an OEM sensor, as supplied by the factory, on a 2002 GMC Safari with a bit less than 125k miles. For this sort of failure, I consider it neither particularly old, nor high-mileage: It was early.

The replacement was cheap and easy to find, and the truck would've run mostly properly it broken, but the point is simple: Just because it's Weatherpak doesn't mean it'll last forever.

From the war stories department, I've seen more than several instances of unfused wire running through a jagged hole in the firewall before being wedged between the battery and the positive cable on a side-terminal GM, to actual working fire trucks with looms of plastic-insulated wire that had lost its plasticity, cracked, and would short randomly with very pretty sparks if wiggled anywhere under the dash.

I once saw an antenna cable catch fire that was routed through the roof of a brand new ambulance. (The ambulance survived, the fire self-extinguished, and the tech responsible learned a few things about assumptions, as did the rest of us.)

Part of my day job involves working with sirens, lights, and communications on public safety vehicles. Much of it is good (or at least safe) work, some of it is excellent, but the rest of the stuff I find is so scary that I cannot (in good conscience) leave it that way even if it's unrelated to the other work that I'm doing. This is the level of incidental repair work where I start taking pictures and documenting every splice, and the customer is always glad to have it fixed and always pays the bill, but...

Back to your main topic (if you're reading this far), what are you using for an ODB-II reader? I've been thinking of putting together some kit (possibly over Bluetooth so it works with both my Droid and laptop) to better understand the workings of that truck, and to diagnose it on the go (if I'm going anywhere far, it's likely for work) and am interested in any opinions or anecdotes you have about specific gear. (And since I'm a sucker for helping folks out, I'll probably end up using it on a lot of different vehicles...)

Re:transcript summary

[Noughmad]

Pilots talk about cheese, the flight attendants ass for 1:59, strange voice yells "alluhu ackbar", tape ends.

So I guess we'll never know what happened!

It was a trap!

Pretty amazing tech

[afidel]

Impact Shock 3400G, 6.5 milliseconds
Penetration Resistance 500 lb. weight from 10 feet
Static Crush 5000 lbs., 5 minutes
High Temperature Fire 1100 C, 30 minutes
Low Temperature Fire 260 C, 10 hours
Deep Sea Pressure and 20,000 feet, 30 days
Sea Water/Fluids Immersion Per ED-56a
The CSMU design has been fully qualified to these requirements and, in fact, exceeds them by considerable margin in key survival areas:
Impact shock has been successfully demonstrated at 4800 G's
High temperature fire exposure has been tested to 60 minutes
Low temperature fire was tested immediately after exposure to 1100 C fire.

From here. Check out the physical design on page 8.

Re:I wonder if NOVA got it right.

[catchblue22]

This was a very interesting documentary. I was particularly interested in the inferences about the user interface approach of Airbus versus Boeing. In short, Airbus planes are controlled with joysticks that translate pilot intentions into actual executable commands to the control surfaces. If the pilot tells the computer to do something stupid, the computer won't do it. Contrast this with Boeing, where the pilots control the plane with a proper control stick that gives more effective feedback to the pilots. In a Boeing airplane, when the computer lowers engine power on autopilot, the engine control lever actually moves in a very visible way. However, on Airbus planes, the levers DO NOT move. The only indication to a pilot that the power has dropped is a small circular readout on a computer screen. The Nova scientists theorized that the pilots didn't realize that the computer had lowered power in anticipation of flying through a thunderstorm, or at least that they realized it too late. They theorize that for about a minute the pilots were flying the plane as if the engines were on high power, when they were actually on a much lower power setting. This, combined with a lack of reliable airspeed data may have caused the pilots to put the plane in an unrecoverable mode of flight. Or maybe it was different. We will know soon enough.

BTW, for those of you outside the US, the above video link won't work. I think the video is on bittorrent somewhere. It is definitely worth watching if you haven't seen it.

Re:I wonder if NOVA got it right.

[catchblue22]

the pilots should still be able to bring the aircraft to area where visual flight rule is possible.

The pilots should be able to fly the plane without airspeed data, according to the Nova documentary. They just set the engine to a particular level and maintain a particular angle of attack. The Nova documentary speculated that due to a variety of factors and distractions that the pilots were unaware of the actual power settings of the airplane. Apparently the airspeed/angle of attack window is quite narrow at that altitude, and if the plane deviates from that window, the airplane may become uncontrollable. It may have taken a brief oversight of the power settings to bring down the plane. Sort of pilot error, perhaps. But there were definitely mitigating factors.

Re:Data can be misleading...

[catchblue22]

Likely as not, nothing will jump out, especially if, say, the autopilot is flying the aircraft using faulty input, e.g., input from an ice-covered sensor. That is likely to cause other sensors to show perverse readings that may (or may not) be very subtle, and may have multiple or ambiguous causes.

Perhaps, but what will be very interesting is the data on the power settings on the airplane, especially in regard to (a) the autopilot reducing power to 70% in anticipation of passing through a thunderstorm and (b) the pilot's changing of that setting to a more appropriate level. The key question is whether or not the autopilot lowered the power before kicking off due to bad airspeed data, and whether or not (and when!) the pilots realized that the power was lower and what they did in response. That goes to the heart of the speculated cause of this crash, according to the Nova episode on the subject.